mental-health-gaming-predictor

Mental Health Gaming Predictor

Predicts depression_score (0–10) based on gaming behavior using a Model Chaining architecture.

Tugas 4: Modeling Experiments — Kelompok 3

Architecture

daily_gaming_hours + competitive_rank
        ↓
   Imputer (predict addiction_level)
        ↓
   [addiction_level, daily_gaming_hours, competitive_rank]
        ↓
      Scaler (StandardScaler)
        ↓
   Predictor (Linear Regression / Random Forest)
        ↓
   depression_score (0–10)

Models

Performance

Linear Regression is the recommended model — nearly identical R² with 70× faster inference.

Risk Classification

Usage

import joblib
import numpy as np
import pandas as pd

# Load models
imputer = joblib.load("model_prediksi_adiksi.pkl")
scaler = joblib.load("scaler_mental_health.pkl")
predictor = joblib.load("model_regresi_mental_health.pkl")

# Input
daily_gaming_hours = 6.0
competitive_rank = 80  # percentile 1-100

# Step 1: Predict addiction level (if unknown)
addiction_level = imputer.predict(pd.DataFrame({
    "daily_gaming_hours": [daily_gaming_hours],
    "competitive_rank": [competitive_rank]
}))[0]

# Step 2: Scale features
features = pd.DataFrame({
    "addiction_level": [addiction_level],
    "daily_gaming_hours": [daily_gaming_hours],
    "competitive_rank": [competitive_rank]
})
scaled = scaler.transform(features)

# Step 3: Predict depression score
score = predictor.predict(scaled)[0]
print(f"Depression Score: {score:.2f}")

Dataset

Synthetic gaming behavior and mental health dataset from Kaggle (968,287 rows, 39 columns). Available at: adeputr4/mental-health-gaming-dataset

Team